One of my concerns with self-driving systems is that they don't have a good model of the world they are operating in. Yes, sure, they build up 3-D models using multiple sensor types, and react to that accordingly, oftentimes with faster reflexes than a human.
However, consider this situation with pedestrians A and B, while driving down a street, approaching an intersection.
Pedestrian A, on the edge of the sidewalk, is faced away from the street, looking at his phone. The chances of him suddenly stepping out in front of my car are exactly 0.0%.
Pedestrian B, on the edge of the sidewalk, is leaning slightly towards the street, and is looking anxiously down the street (the wrong way). The chances of B suddenly stepping out is high, and the chances of B not seeing my car are very high (because B is being stupid/careless, or else is used to UK traffic patterns).
I, as a human, will react to those situations differently, preemptively slowing down if I recognize a situation like with Pedestrian B. Autonomous driving systems will treat both situations exactly the same.
From everything I've read about so far, current driving systems have no way of distinguishing between those two specific situations, or dealing with the myriad of other similar issues that can occur on our streets.
The cynic in me thinks self-driving will require the same, the rules and conservative practices will only come after a bunch of highly publicized, avoidable failures that leave their own trail of blood.
This is not a field where 'move fast and break stuff' is a mantra worth anything at all, the stuff that is broken are people's lives and as such you'd expect a far more conservative attitude, akin to what we see in Aerospace and medical equipment software development practices.
But instead barely tested software gets released via over-the-air updates and it's anybody's guess what you'll be running the next time you turn the key.
I agree with you that the software should have been able to detect that something was wrong either way, either it was already halfway in the wrong lane or it was heading to be halfway in the wrong lane, a situation that should not have passed without the car at least alerting the driver.
And from what we have seen in the video it just gave one inferred situation priority over the rest and that particular one happened to be dead wrong about it's interpretation of the model.